Crypto phishing alerts: what Bitget’s 7.4x rise means11 min read2,259 words

Crypto phishing alerts: what Bitget’s 7.4x rise means

Bitget Wallet flagged 37,000 weekly signing requests for phishing in Q3, up 7.4x. Understand what the data shows and how to assess crypto wallet prompts.

crypto phishingBitget Wallet securitywallet signing requestsAI crypto threat detectioncrypto wallet safety
Crypto phishing alerts: what Bitget’s 7.4x rise means

Bitget's 7.4x Alert Surge: What Crypto Users Should Know

A wallet warning is only helpful if you understand what it's measuring. Bitget Wallet says its average weekly count of signing requests flagged as potential phishing jumped from about 5,000 in Q2 2026 to around 37,000 in Q3 - roughly a 7.4-fold increase. That's a big shift in what the wallet flags, but it doesn't automatically mean phishing attacks or user losses rose by the same amount.

That matters because Bitget rolled out AI-assisted risk detection in Q2. More alerts can come from more dangerous activity, better detection, or both - and Bitget says it can't break down how much each factor contributed. Also, the metric counts flagged signing interactions, not unique attacks, victims, or confirmed losses.

So what do the numbers tell you - and what don't they? This guide walks through what the alert counts mean, how AI-style checks can help catch risky wallet behavior, and how to review a signing prompt before you approve anything - whether you're swapping tokens, connecting to a DeFi app, or authorizing a smart contract.

What does Bitget Wallet's 7.4x increase measure?

Bitget Wallet reported that it flagged an average of about 37,000 signing requests per week for potential phishing in Q3 2026, compared with 5,000 per week in Q2. Simple math puts the Q3 figure at about 7.4 times the Q2 figure, or about a 640% increase over the earlier weekly average.

But the key unit is flagged signing requests. A signing request is an interaction that asks a wallet user to authorize or sign something - like a transaction, a token approval, or another message. This number isn't the same thing as a count of distinct phishing campaigns. One campaign can trigger many requests, and one user can see several. And a flagged request still isn't proof that the user signed it or lost funds.

Bitget Wallet's chief operating officer Alvin Kan told Cointelegraph that the increase likely reflects both more threat activity and improved detection. He also said the data doesn't show how much each contributed. That rules out a casual conclusion like "phishing grew 7.4x." The metric reflects what the wallet detected - not the full volume of malicious activity across crypto.

There's also a timing and method wrinkle. The comparison runs from Q2 to Q3, and Bitget introduced AI-assisted risk detection in Q2. If the wallet's coverage improves, it can flag more suspicious interactions even if the underlying threat level stays the same. On the other hand, a real increase in phishing attempts could also raise alerts even if detection quality didn't change much.

A useful distinction: activity, detection and harm

When you look at a security metric, it helps to separate three questions:

  • Activity: How many suspicious attempts happened?
  • Detection: How many did the product flag?
  • Harm: How many users signed, and what losses followed?

Bitget's weekly alert figure mostly answers the detection question. It doesn't tell you the other two by itself. That doesn't make it useless - it shows the scale of activity reaching the wallet's warning system and suggests the company is investing in monitoring - but it shouldn't be treated as a market-wide attack rate or a direct measure of financial damage.

If you want a clearer picture, you'd need follow-up details like: how many unique domains or campaigns were involved, what share of alerts were later confirmed malicious, how many warnings users actually saw, and how many potential losses were prevented versus completed. Without those pieces, alert counts are best treated as a signal - not a full risk report.

Why might phishing alerts rise so quickly?

An alert total depends on two things: what's happening out there (the environment) and how the detector works. Change either one, and the number can jump.

Bitget points to both more threat activity and improved detection, while saying it can't separate their effects. The detector did change during this comparison period. Bitget introduced AI-assisted risk detection in Q2 and says it expanded those checks. When a system looks at more signals - or finds patterns it previously missed - it can flag more signing requests even if fewer attacks are "actually" succeeding.

That's a common interpretation problem in security operations: more reports can mean visibility improved, not that defenses failed.

The underlying activity may also have changed. Phishing attempts can arrive through fake sites, messages, or interfaces designed to look like familiar wallet providers. A user may then be nudged toward a wallet action that looks routine but results in permissions or approvals with consequences that are hard to undo. The reporting around this broader risk includes examples such as a trader reportedly losing nearly $1 million after signing a malicious Ethereum token approval in July, and an August Rapid7 investigation describing a campaign targeting 885,000 phone numbers with websites impersonating crypto wallet providers.

Those examples show different stages of the same pipeline: a message or website gets a user to a malicious page; that page then tries to prompt a signature or approval; the wallet may detect something suspicious, but the user still decides whether to proceed. A bigger alert count doesn't tell you how often the attempt reached each stage.

What a rising count can - and cannot - tell you

A rising alert count can be a useful operational signal. It suggests the wallet is encountering or detecting more interactions it considers risky. It may justify stronger monitoring, clearer warnings, and extra attention to approvals. But it's not enough to establish a trend in total crypto crime, compare platforms in a fair way, or estimate the odds that any one user gets targeted.

To do those comparisons, you'd need consistent definitions, consistent reporting periods, and comparable counting methods. One wallet might issue more alerts because it scans more aggressively; another might issue fewer because it sees less activity or misses more threats. Alert volume needs context before it becomes a measure of risk.

How can AI help detect risky wallet activity?

Bitget Wallet says its expanded AI checks examine website code and domain registration details for signs of fraud. It also says the wallet simulates transactions to spot hidden permissions before a user signs. Those checks target two different parts of the decision: whether the site itself looks suspicious, and what the wallet action could allow.

That mix matters because a convincing page doesn't automatically mean the interaction is safe. A fraudulent site can copy a legitimate service, and a transaction request can include permissions that aren't obvious from what the user sees. Checking both the source of the request and the likely effect of the signature gives the wallet more context than relying only on a domain blocklist.

Bitget also reported that AI helped identify 90% of newly blocked risky domains in October, compared with 36% in Q2. It further says nearly 70% of the domains flagged in October didn't show up in the external threat feeds it checked. These are Bitget's figures, and the comparison is narrow: they describe Bitget's detection workflow and how it compares to those feeds. They don't prove that every flagged domain was malicious, or that Bitget's approach beats every alternative system.

Why threat feeds are not enough

Threat feeds can help by listing known domains and previously reported issues. But a domain that doesn't show up in a feed isn't automatically safe - it might be new, under-reported, or simply missing from what the feed covers. Bitget's point that many flagged domains didn't match feed alerts highlights why adding other signals can matter, instead of trusting only existing lists.

AI can help process those signals at scale, but it's not a guarantee. Detection depends on input quality, the rules used to classify risk, and how the system handles unfamiliar cases. A warning can be wrong or incomplete, and some risky interactions will still slip through. In practice, a good warning is an extra layer of review - not a substitute for understanding what you're signing.

For wallet providers, the operational question isn't just how many domains an AI system flags. It's whether warnings are accurate, timely, and understandable - and whether users can tell why an action is risky. High alert volume without clear explanations can lead to warning fatigue. Fewer alerts can still be better if they clearly explain the permission involved. Detection and communication have to work together.

How should you assess a wallet signing request?

Treat a wallet prompt like a request for authorization, not like a routine pop-up you can dismiss. The exact meaning depends on the transaction, network, and application, but a few checks catch a lot of avoidable mistakes. Slow down when the request looks out of place, when you end up routed through an unfamiliar site or flow, or when the warning doesn't match what you intended to do.

A practical pre-signing check

Before you approve, go through this:

  1. Confirm the action. Did you initiate this transaction or message? If you didn't expect a prompt, reject it and verify using a route you already trust.
  2. Check the site and context. Look closely at the domain and think about how you arrived there. A familiar layout isn't verification, and an unflagged domain isn't proof of safety.
  3. Read what the wallet is telling you. If it points to a suspicious site, unusual transaction, or risky permission, don't dismiss it just because the page offers a reward or claims an urgent "fix."
  4. Understand the permission. For token approvals, check which asset is involved and whether the permission looks broader or longer-lasting than what the action really needs. If you can't understand it from the wallet's explanation, don't guess.
  5. Stop when the request is unclear. Legit services shouldn't require you to sign something you can't reasonably explain. Verify before continuing.

This approach is intentionally conservative. Rejecting something legitimate can cost you time or break a flow, but signing a malicious approval can put assets at risk - and blockchain actions may be difficult or impossible to reverse. When the cost of pausing is low and the potential downside is high, waiting is usually the smarter move.

For teams building wallets or DeFi apps, the same principle applies at the product level: make the effect of a signature clear in plain language, separate "connection" from "approval," and surface risky permissions before the user signs. Warnings should explain what's concerning and what the user can do instead. If every alert looks equally urgent, users learn to ignore them.

Key takeaways

  • Bitget's reported 7.4-fold increase is for weekly flagged signing requests, not confirmed attacks or losses.
  • Improved monitoring may explain part of the change; Bitget says it can't quantify how much detection versus threat activity drove the numbers.
  • AI-assisted checks can add context from websites, domain details, and transaction simulations, but they can't guarantee an interaction is safe.
  • Read the requested action and permission before signing. If you can't explain what you're authorizing, stop.

Conclusion

Bitget Wallet's Q3 reporting shows a sharp rise in signing requests its system flagged for potential phishing. That doesn't prove that phishing attacks or losses increased by the same multiple. Bitget introduced AI-assisted detection in Q2, and the company says the increase likely reflects better monitoring alongside changes in threat activity.

The main lesson is to avoid treating a single alert number as the full picture of crypto security. Check what's being counted, how the detection changed, and whether the figures connect to confirmed harm. For users, the takeaway is straightforward: treat every signature as authorization, read what the wallet says the signature will do, and pause when the request isn't clear. That habit still matters, even as detection systems get better.

Frequently Asked Questions

Not according to the published figure. Bitget Wallet reported a 7.4-fold increase in weekly signing requests flagged as potential phishing. The company said the rise likely reflects both threat activity and improved detection, and it could not quantify each factor. The metric is not a count of confirmed attacks or financial losses.
It is a request for a wallet user to authorize or sign an interaction, such as a transaction, token approval or message. The consequences depend on what is being signed. Before approving, check the wallet’s explanation and make sure the requested action matches what you intended to do.
No. A warning is a risk signal, not a verdict that provides complete context. Likewise, the absence of a warning does not establish that a site or transaction is safe. Use the wallet’s explanation as one input, then verify the site and understand the permission being requested.
Do not sign it just to make the prompt disappear. Reject the request, revisit the action you were trying to complete, and check whether the site and permission make sense. If you cannot verify what the approval allows, stop rather than relying on a guess.

Related articles